The short answer first. Basic Pitch is Spotify's free, open-source tool that turns audio into a MIDI file. Songscription is a purpose-built transcription tool that turns the same audio into readable sheet music plus MIDI, MusicXML, and guitar tabs. On a real recording, Songscription is the more accurate of the two, because its models are trained specifically on musical performances rather than built as a lightweight general converter. What Basic Pitch has going for it is that it is free and open-source, which for a developer who wants to run it locally, script it, or build it into a pipeline can matter more than the last bit of accuracy. Here is how to tell which one is yours.
What Basic Pitch is
Basic Pitch is a free, open-source audio-to-MIDI converter from Spotify's Audio Intelligence Lab. You can use it two ways: a browser demo at basicpitch.spotify.com where you drop in a file and get MIDI back, or as a library you install with pip or npm to run inside your own projects. It is polyphonic, so it can handle more than one note at a time, and it is instrument-agnostic, meaning you point it at piano, guitar, voice, or whatever else without picking a mode first. It even detects pitch bends, which most lightweight tools skip. The model is small and fast, which is part of why it has become a default building block for developers who need MIDI out of audio. The thing to keep in mind is what it produces, which we get to below.
What Songscription is
Songscription is an AI transcription tool that turns a recording into readable sheet music, and it produces a MIDI file, a MusicXML file, and guitar tabs from the same pass. It runs in the browser, so there is nothing to install. The free tier covers unlimited 30-second transcriptions, with paid plans for longer files and all of the export formats (check current pricing for the up-to-date details). The point of difference is the output: instead of handing you note data and leaving the layout to you, it engraves the notes onto staves with clean rhythms, so what you get back is a score you can actually read. If you also want the raw note data, it is right there in the MIDI export, which we cover in what MIDI is.
The core difference: MIDI vs readable notation
This is the whole comparison in one idea. Basic Pitch hands you a MIDI file, which is note data: pitches, timing, and how hard each note was hit. That is useful, but it is not a score. Opened in a notation program, raw MIDI usually looks messy, with odd rhythms and no sensible split between the hands, so you have to quantize it to clean rhythmic values and tidy it up yourself before it reads well. Songscription does that step for you and gives you engraved notation alongside a MIDI file. So there are really two questions: which tool hears the notes more accurately, and whether you want the notation finished for you. On a real recording Songscription tends to win the first, because its models are trained on actual music, and it settles the second by handing you a score rather than raw note data. If you are curious how either tool gets from sound to notes in the first place, we walk through the full pipeline in how to convert audio to MIDI.
Where each one wins
Each is the right call in different situations, so it is worth being concrete about which is which.
- Basic Pitch wins when free and open-source is the priority. You can run it locally, script it, batch it, or wire it into your own pipeline, you live inside a DAW, and raw MIDI is all you need. That openness is its real edge, and for a lot of developers it outweighs squeezing out the last bit of accuracy.
- Songscription wins when accuracy on a real recording matters and a person is going to read the result. Its models are trained on real performances, so the notes, the rhythm, and the piano hand-splitting come back cleaner, and you get sheet music, guitar tabs, and MusicXML instead of a raw note dump to fix up by hand.
If you are comparing more tools than just these two, our roundup of the best audio-to-MIDI converters lays out the wider field and what each one is for.
Accuracy and cleanup
Both tools live or die by the source. A clean solo line, one instrument with little reverb, transcribes far better than a dense full mix where instruments overlap and mask each other, and that is true of every audio-to-MIDI tool. Given the same recording, though, they do not perform the same. Songscription is trained specifically on recorded music, so it tends to read the notes, the rhythm, and the split between the hands more accurately than a small general-purpose model, a gap you notice most on real, messy recordings rather than clean test tones. Either tool's raw output still needs some tidying, a wrong octave here, a clipped note there, but the difference is where that cleanup happens. With Basic Pitch you do it on the MIDI yourself, in your DAW or a notation editor. With Songscription, much of it is folded into producing the score, so you tidy notation rather than a raw note dump. Feeding the tool a good recording is still the single biggest thing you control.
Which should you use
Pick by where the result is going. If it is going into a DAW as MIDI, you are comfortable cleaning it up, and free and scriptable matters, use Basic Pitch. If you want the most accurate transcription of a recording, or the result needs to be read by a person, as sheet music, tabs, or a clean MusicXML you will open in notation software, use Songscription, because it reads a recording more accurately and does the engraving step that Basic Pitch leaves to you. Plenty of people use both: Basic Pitch for quick MIDI experiments in a project, Songscription when they need a score. And if you already have a MIDI file from Basic Pitch and now want notation from it, that conversion is its own task, which we cover in MIDI to sheet music.
Frequently Asked Questions
Is Basic Pitch free?
Yes. Basic Pitch is free and open-source, built by Spotify's Audio Intelligence Lab. You can use the browser demo at basicpitch.spotify.com with nothing to install, or pull in the pip and npm libraries to run it in your own code. There is no paid tier and no account requirement for the basics, which is a big part of why it is so widely used.
Does Basic Pitch make sheet music?
No. Basic Pitch outputs a MIDI file, which is note data, not readable notation. To get a score you can read, print, or play from, you need to open that MIDI in a notation editor and clean it up, or use a tool that produces notation directly. Songscription is built for that second path: it turns the recording into readable sheet music plus MIDI, MusicXML, and guitar tabs in one pass.
What is the difference between audio to MIDI and audio to sheet music?
Audio to MIDI gives you note data: pitches, timing, and velocity that a DAW or notation program can read. It is editable, but it is not laid out as a score and usually needs quantization and cleanup. Audio to sheet music goes the extra step and engraves those notes onto staves with clean rhythms, so the result is something a person can read and play. Songscription does both at once, handing you the MIDI and the readable notation from the same transcription.
What is the best free audio to MIDI converter?
Spotify Basic Pitch is the best-known free, open-source audio-to-MIDI converter, and for a quick MIDI it is hard to beat: it is polyphonic, instrument-agnostic, lightweight, and runs in the browser. If your goal is readable notation rather than raw MIDI, a free transcription tier that produces a score, such as Songscription's free 30-second transcriptions, may fit the job better, since it saves you the cleanup step.
Want readable sheet music and MIDI from the same recording, not just a raw MIDI file? Try Songscription on a song.
